refine
Skill5 starsChanged 44 days agoMITView on GitHub
User-invoked only. Pre-implementation refinement: clarifying questions, ranked approaches, posted back to the ticket
- Reads credentials
44 other projects use a version of this.
Adopted by
Most recently taken up by
- kbhuwalka/skillstheir own version, first seen Nov 15, 2026
- automagik-si/genietheir own version, first seen Nov 15, 2026
- ghostdogpr/skillstheir own version, first seen Nov 13, 2026
- ryanportfolio/winuse-mcptheir own version, first seen Nov 8, 2026
- ryanportfolio/Harness-Firmwaretheir own version, first seen Nov 7, 2026
- ryanportfolio/HarnessFirmware.comtheir own version, first seen Nov 5, 2026
Used with
Projects with it also use addskill, apply-firmware and 10 more.
Other files people use alongside this one.
Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
ryanportfolioSkilladdskill
Create, import, update, or install repo or personal skills: runtime ownership, resources, discovery checks.
Also in 6 projects that use this
ryanportfolioSkillapply-firmware
Bring current folder (plain folder, existing repo, old or partial firmware copy) up to full Harness Firmware in place. Mirror external repo → adopt-repo.
Also in 6 projects that use this
- +10 more from ryanportfolio
Other versions44 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 44
- PrimeIntellect-ai/prime-agent★ 23ksame name
- windmill-labs/windmill★ 19ksame name
- DAIR-AutoResearch/AutoSci★ 1.7ksame name
- PKU-DAIR/AutoSci★ 1.7ksame name
- skyllwt/AutoSci★ 1.7ksame name
- skyllwt/OmegaWiki★ 1.7ksame name
- rsi-gear/gear★ 422same name
- AlexPEClub/ai-coding-starter-kit★ 382same name
- automagik-dev/genie★ 353same name
- automagik-si/genie★ 353same name
34 more not listed
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository24 other files it gives its agents.
- AGENTS.mdAGENTS.md
- assign-epicSkill
- briefingSkill
- bump-versionSkill
- captureSkill
- consultantSkill
- dependabot-reviewSkill
- fix-cicdSkill
- log-timeSkill
- morlockSkill
- morningSkill
- multi-repoSkill
- nightly-checkSkill
- pickupSkill
- pr-action-review-mine-loopSkill
- pr-action-reviewSkill
- pr-choreSkill
- pr-review-loopSkill
- prSkill
- pushSkill
- qa-review-actionSkill
- runSkill
- syncSkill
- wrap-upSkill
The file
We split this skill into the rules it gives an agent
6 rules. 0 are common, 3 are distinctive, and 3 are very specific to this project.
0
Common
Most of the 38 edited versions we read keep these.
3
Distinctive
This author's own choices. Worth a look.
3
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 9
Ticket input: use the value supplied with the skill invocation.
- Very specific: line 15
If the ticket input is a bare number (e.g.
- Very specific: line 19
Check that JIRA_BASE_URL, JIRA_API_TOKEN, JIRA_EMAIL, and JIRA_ACCOUNT_ID are set in .env.
- Very specific: line 57
If the ticket has a fields.parent, fetch that ticket to understand the broader story or epic:
- Distinctive: line 93
Ask clarifying questions to the user until you have a complete understanding of the ticket.
- Distinctive: line 154
After posting, pause and ask: "Refinement posted — do you want to adjust anything before we plan the implementation?"
“Common” is measured: kept by at least 6 in 10 of the 38 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Review a Jira ticket by ID: surface clarification questions, propose implementation approaches in priority order, then post conclusions as a comment on the ticket.
Ticket input: use the value supplied with the skill invocation.
Step 0 — Normalise the ticket ID
If the ticket input is a bare number (e.g. 25), prepend $JIRA_PROJECT_KEY- to get e.g. PROJ-25. Store the result as TICKET_ID and use it for all curl commands below.
Step 1 — Verify credentials
Check that JIRA_BASE_URL, JIRA_API_TOKEN, JIRA_EMAIL, and JIRA_ACCOUNT_ID are set in .env. If any are missing, stop and ask the user to add them. Source .env before all Jira API calls.
Step 2 — Fetch the ticket and its comments
source .env && curl -s -u "$JIRA_EMAIL:$JIRA_API_TOKEN" \
"$JIRA_BASE_URL/rest/api/3/issue/$TICKET_ID?expand=subtasks,renderedFields"Parse the response and extract:
fields.summaryfields.description(full rendered text)fields.assigneefields.status.namefields.issuetype.namefields.parent(if present — epic or story)fields.subtasks(array of subtask keys)fields.issuelinks(linked issues)fields.labelsfields.priority.namefields.customfield_10014(epic link, if present)
Then fetch the comment thread, ordered oldest-first:
source .env && curl -s -u "$JIRA_EMAIL:$JIRA_API_TOKEN" \
"$JIRA_BASE_URL/rest/api/3/issue/$TICKET_ID/comment?orderBy=created"For each comment, extract author.displayName, created, and the plain-text body. Comments represent decisions and context that already exist on the ticket and take precedence over any fresh analysis — read them all before proposing anything.
Step 3 — Check assignee and status
If the ticket is already assigned or in development, ask the user to confirm if they still want to refine it. Remember to mention the JIRA user that the ticket is assigned to, the current status and how many days it has been in that status.
Step 4 — Fetch parent/epic and subtask context
If the ticket has a fields.parent, fetch that ticket to understand the broader story or epic:
source .env && curl -s -u "$JIRA_EMAIL:$JIRA_API_TOKEN" \
"$JIRA_BASE_URL/rest/api/3/issue/<parent-key>?expand=renderedFields"For each key in fields.subtasks, fetch the full issue:
source .env && curl -s -u "$JIRA_EMAIL:$JIRA_API_TOKEN" \
"$JIRA_BASE_URL/rest/api/3/issue/<subtask-key>?expand=renderedFields"Step 5 — Read relevant codebase context
Based on the ticket's domain (frontend, backend, infrastructure, shared), read the relevant documentation and source files to understand the current state before forming opinions:
- Always:
CONTRIBUTING.md,docs/development/engineering-standards.md - Architecture changes:
docs/architecture/context.md,docs/architecture/containers.md,docs/adr/ - Frontend work:
docs/development/react-conventions.md,apps/web/ - Backend work:
docs/development/backend-patterns.md,apps/api/ - Quality/testing:
docs/development/quality-strategy.md
Step 6 — Analyse the ticket
With all context gathered, think through:
- Clarity — Is the acceptance criteria precise and testable? Are there ambiguous terms, missing edge cases, or unstated assumptions?
- Scope — Is the scope appropriate? Could it be split? Does it implicitly require other changes not mentioned?
- Constraints — What architectural, performance, security, or accessibility constraints apply? Which documented patterns are relevant?
- Approaches — What are the distinct ways this could be implemented? Consider trade-offs in complexity, testability, reversibility, and alignment with existing patterns.
- Discussion – Read the comments. If there are any definitive answers by human users (not actioned by an AI coding agent), those take precedence.
Step 7 — Ask clarifications
- Ask clarifying questions to the user until you have a complete understanding of the ticket. For each question, reference the specific part of the ticket that is unclear. If the ticket is already clear, explicitly state that no clarifications are needed.
- If there are distinct alternatives to be considered, ask the user to select one. Present up to 2 pros and cons of each alternative, and which ones you prefer and why. Highlight your preference in the name/id of each option and use 1-5 stars to rate each approach.
Step 8 — Produce the refinement brief
Output a structured brief:
## Refinement: <ticket-id> — <summary>
### Clarifications needed
List questions where the ticket is ambiguous, incomplete, or where assumptions need validating.
Be specific — reference the exact part of the description or acceptance criteria that is unclear.
If the ticket is clear, say so explicitly rather than inventing questions.
### Proposed approaches
**1. <Preferred approach — recommended>**
- What: concise description of the approach
- Why preferred: alignment with existing patterns, testability, simplicity, reversibility
- Trade-offs: what you give up or accept
- Risks: what could go wrong
**2. <Alternative approach>**
- What: concise description
- Why it was considered: a legitimate reason it's viable
- Trade-offs: why it ranks lower
*(Add further alternatives only if genuinely distinct — do not pad)*
### Recommendation
One paragraph synthesising the preferred path and why, referencing the codebase context read in Step 5.
### Open questions before implementation
Numbered list of things that must be resolved before coding starts. Separate from clarifications — these are architectural or risk decisions, not ambiguities in the ticket text.
## Clarifications resolved
- Explain that these were the clarifications that were resolved by the user when asked by the coding agent.
- For every question that was clarified, list the question and the answer that was provided by the user. This is a record of how the ticket evolved from its original state to the clarified state.
- Do not summarise the questions or the answers, this is an audit trail.
- Do not use block quotes, they break the ADF format.
Step 9 — Post conclusions as a Jira comment
Format the brief above as Atlassian Document Format (ADF) JSON and post it as a comment on the ticket:
source .env && curl -s -u "$JIRA_EMAIL:$JIRA_API_TOKEN" \
-X POST "$JIRA_BASE_URL/rest/api/3/issue/$TICKET_ID/comment" \
-H "Content-Type: application/json" \
-d '<ADF comment body>'The comment body must use ADF. Structure it with headings for each section (Clarifications, Approaches, Recommendation, Open questions).
Confirm to the user that the comment was posted successfully (HTTP 201) or report the error if it failed.
After posting, pause and ask: "Refinement posted — do you want to adjust anything before we plan the implementation?"
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

